An Analysis of Protein Patterns Present in the Saliva of Diabetic Patients Using Pairwise Relationship and Hierarchical Clustering.

Molecular diagnosis is based on the quantification of RNA, proteins, or metabolites whose concentration can be correlated to clinical situations. Usually, these molecules are not suitable for early diagnosis or to follow clinical evolution. Large-scale diagnosis using these types of molecules depends on cheap and preferably noninvasive strategies for screening. Saliva has been studied as a noninvasive, easily obtainable diagnosis fluid, and the presence of serum proteins in it enhances its use as a systemic health status monitoring tool. With a recently described automated capillary electrophoresis-based strategy that allows us to obtain a salivary total protein profile, it is possible to quantify and analyze patterns that may indicate disease presence or absence. The data of 19 persons with diabetes and 58 healthy donors obtained by capillary electrophoresis were transformed, treated, and grouped so that the structured values could be used to study individuals’ health state. After Pairwise Relationships and Hierarchical Clustering analysis were observed that amplitudes of protein peaks present in the saliva of these individuals could be used as differentiating parameters between healthy and unhealthy people. It indicates that these characteristics can serve as input for a future computational intelligence algorithm that will aid in the stratification of individuals that manifest changes in salivary proteins.